Modem database applications are typically designed using relational data models that are implemented by well known technologies such as DB2. For many applications, a relational data model works well. For some applications, however, the relational database model limits the ability to represent complex conceptual information. Such systems typically store data in tables that include one or more columns. When data is stored in the tables, the data may not occupy all or a portion of one or more columns. Therefore, data storage space that includes unused columns may be unused. Additionally, such systems typically require a new table schema to be defined for each data collection to be saved. Therefore, data to be stored must be assembled by the system to populate the table's schema within the table. Additionally, systems typically only allow navigation through the table in a downward direction.
These tables contain named columns of information and one row for each data observation. Related tables may be joined by one or more columns that contain a common field. The descriptions of all tables, columns, and relationships are contained in the database schema, which is typically maintained by a full-time database administrator (DBA). The definition of new tables, columns, or relationships usually requires the intervention of a DBA.
Many applications typically store dense lists of information, such as financial data or employee records, where most data columns are fully populated. Other applications, such as name and address directories or product catalogues, contain sparse information. That is, out of many potential data elements, only a few may exist in any particular record. Address directories and product catalogues typically maintain relationships, especially inheritances that are usually more easily understood when expressed in hierarchical trees rather than joined relational tables.
Another drawback of existing systems is that data collections to be stored must be transformed to a format compatible with a table to be used to store the data. This may require modification of existing schema or reformatting a database to a particular format. This may include reorganization and possible downtime for the database.
Another drawback with existing systems is that multiple disk reads may be necessary for retrieving rows within a table. For example, if a user desires to further limit a search, and the user inserts additional criteria for searching an object, existing systems typically access a server each time an additional search criteria is input. This may slow the performance of the server and increase processing time.
Relational databases are used to store relationships among tables. Relational databases, however, are not very suitable for storing hierarchies. Therefore, hierarchical databases typically are used to store hierarchies. Data elements are stored as fixed lengths. If a data element does not occupy the fixed lengths, the excess data storage space may be wasted.
Existing systems are typically slow to resolve queries. For example, a query may take one hundred (100) milliseconds to be resolved. Another drawback is that applications are typically coded to retrieve one or at most a few attributes on any one server call. This may lead to multiple calls being issued to a server against the same data collection. Furthermore, because only portions of a data collection are retrieved, existing systems typically parse the data collection for particular information being requested. Parsing may occur for each call issued to a server. This may also increase processing time.
These and other drawbacks exist.